In [1]:
# Import all the libraries
import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)

import matplotlib.pyplot as plt#visualization
from PIL import  Image
%matplotlib inline


import seaborn as sns#visualization
import itertools
import warnings
warnings.filterwarnings("ignore")
from plotly.offline import iplot
import plotly.graph_objs as go#visualization
import plotly.subplots as sp #visualization
import plotly.figure_factory as ff#visualization

Data Import¶

In [2]:
bank = pd.read_csv('Bank Customer Churn Prediction.csv')

Data Overview¶

In [3]:
bank.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 10000 entries, 0 to 9999
Data columns (total 12 columns):
 #   Column            Non-Null Count  Dtype  
---  ------            --------------  -----  
 0   customer_id       10000 non-null  int64  
 1   credit_score      10000 non-null  int64  
 2   country           10000 non-null  object 
 3   gender            10000 non-null  object 
 4   age               10000 non-null  int64  
 5   tenure            10000 non-null  int64  
 6   balance           10000 non-null  float64
 7   products_number   10000 non-null  int64  
 8   credit_card       10000 non-null  int64  
 9   active_member     10000 non-null  int64  
 10  estimated_salary  10000 non-null  float64
 11  churn             10000 non-null  int64  
dtypes: float64(2), int64(8), object(2)
memory usage: 937.6+ KB
In [4]:
print ("Rows     : " ,bank.shape[0])
print ("Columns  : " ,bank.shape[1])
print ("\nFeatures : \n" ,bank.columns.tolist())
print ("\nUnique values :  \n",bank.nunique())

Rows     :  10000
Columns  :  12

Features : 
 ['customer_id', 'credit_score', 'country', 'gender', 'age', 'tenure', 'balance', 'products_number', 'credit_card', 'active_member', 'estimated_salary', 'churn']

Unique values :  
 customer_id         10000
credit_score          460
country                 3
gender                  2
age                    70
tenure                 11
balance              6382
products_number         4
credit_card             2
active_member           2
estimated_salary     9999
churn                   2
dtype: int64
In [5]:
bank.isnull().sum()
Out[5]:
customer_id         0
credit_score        0
country             0
gender              0
age                 0
tenure              0
balance             0
products_number     0
credit_card         0
active_member       0
estimated_salary    0
churn               0
dtype: int64

Replace all the blank values with NAN¶

In [6]:
bank = bank.replace(" ", np.NaN)
bank.head()
Out[6]:
customer_id credit_score country gender age tenure balance products_number credit_card active_member estimated_salary churn
0 15634602 619 France Female 42 2 0.00 1 1 1 101348.88 1
1 15647311 608 Spain Female 41 1 83807.86 1 0 1 112542.58 0
2 15619304 502 France Female 42 8 159660.80 3 1 0 113931.57 1
3 15701354 699 France Female 39 1 0.00 2 0 0 93826.63 0
4 15737888 850 Spain Female 43 2 125510.82 1 1 1 79084.10 0
In [7]:
# check for isnull() again
bank.isnull().sum()
Out[7]:
customer_id         0
credit_score        0
country             0
gender              0
age                 0
tenure              0
balance             0
products_number     0
credit_card         0
active_member       0
estimated_salary    0
churn               0
dtype: int64

Data Cleaning¶

We will keep 11 input variables and 1 output variable (Churn) to create the model¶

  • customer_id
  • credit_score
  • country
  • gender
  • age
  • tenure
  • balance
  • products_number
  • credit_card
  • active_member
  • estimated_salary
  • churn
In [8]:
missing_data = bank.isnull()
missing_data
Out[8]:
customer_id credit_score country gender age tenure balance products_number credit_card active_member estimated_salary churn
0 False False False False False False False False False False False False
1 False False False False False False False False False False False False
2 False False False False False False False False False False False False
3 False False False False False False False False False False False False
4 False False False False False False False False False False False False
... ... ... ... ... ... ... ... ... ... ... ... ...
9995 False False False False False False False False False False False False
9996 False False False False False False False False False False False False
9997 False False False False False False False False False False False False
9998 False False False False False False False False False False False False
9999 False False False False False False False False False False False False

10000 rows × 12 columns

In [9]:
# Count missing values
for column in missing_data.columns.values.tolist():
    print(column)
    print (missing_data[column].value_counts())
    print("") 

customer_id
False    10000
Name: customer_id, dtype: int64

credit_score
False    10000
Name: credit_score, dtype: int64

country
False    10000
Name: country, dtype: int64

gender
False    10000
Name: gender, dtype: int64

age
False    10000
Name: age, dtype: int64

tenure
False    10000
Name: tenure, dtype: int64

balance
False    10000
Name: balance, dtype: int64

products_number
False    10000
Name: products_number, dtype: int64

credit_card
False    10000
Name: credit_card, dtype: int64

active_member
False    10000
Name: active_member, dtype: int64

estimated_salary
False    10000
Name: estimated_salary, dtype: int64

churn
False    10000
Name: churn, dtype: int64

Dependent Variable or Variable of Interest is Churn¶

In [10]:
#Separating churn and non churn customers

churn     = bank[bank["churn"] == 1]
not_churn = bank[bank["churn"] == 0]

Customers who stay and left¶

In [11]:
print ('Customers who stayed', len(not_churn))
print ('Customers who left', len(churn))

Customers who stayed 7963
Customers who left 2037
In [12]:
#Separating catagorical and numerical columns
Id_col     = ['customer_id']
target_col = ["churn"]
cat_cols   = bank.keys().tolist()
In [13]:
cat_cols
Out[13]:
['customer_id',
 'credit_score',
 'country',
 'gender',
 'age',
 'tenure',
 'balance',
 'products_number',
 'credit_card',
 'active_member',
 'estimated_salary',
 'churn']
In [14]:
cat_cols   = bank.nunique()[bank.nunique() < 6].keys().tolist()
In [15]:
cat_cols
Out[15]:
['country',
 'gender',
 'products_number',
 'credit_card',
 'active_member',
 'churn']
In [16]:
cat_cols   = [x for x in cat_cols if x not in target_col]
num_cols   = [x for x in bank.columns if x not in cat_cols + target_col + Id_col]
In [17]:
cat_cols
Out[17]:
['country', 'gender', 'products_number', 'credit_card', 'active_member']
In [18]:
num_cols
Out[18]:
['credit_score', 'age', 'tenure', 'balance', 'estimated_salary']
In [19]:
Id_col
Out[19]:
['customer_id']
In [20]:
target_col
Out[20]:
['churn']

Exploratory Data Analysis¶

Churn No-Churn Ratio¶

In [21]:
#labels
lab = bank["churn"].value_counts().keys().tolist()
#values
val = bank["churn"].value_counts().values.tolist()

trace = go.Pie(labels = lab ,
               values = val ,
               marker = dict(colors =  [ 'royalblue' ,'lime'],
                             line = dict(color = "white",
                                         width =  1.3)
                            ),
               rotation = 90,
               hoverinfo = "label+value+text",
               hole = .5
              )
layout = go.Layout(dict(title = "Churn vs. Non-Churn",
                        plot_bgcolor  = "rgb(243,243,243)",
                        paper_bgcolor = "rgb(243,243,243)",
                       )
                  )

data = [trace]
fig = go.Figure(data = data,layout = layout)

iplot(fig)

From the above graph, majority of the people(79.6%) still continue the service of the bank.¶

In [22]:
data
Out[22]:
[Pie({
     'hole': 0.5,
     'hoverinfo': 'label+value+text',
     'labels': [0, 1],
     'marker': {'colors': ['royalblue', 'lime'], 'line': {'color': 'white', 'width': 1.3}},
     'rotation': 90,
     'values': [7963, 2037]
 })]
In [23]:
bank['churn'].value_counts()
Out[23]:
0    7963
1    2037
Name: churn, dtype: int64

Varibles distribution in Churn and Non-churn category: Pie Charts for Categorical variables¶

In [24]:
def plot_pie(column) :
    
    trace1 = go.Pie(values  = churn[column].value_counts().values.tolist(),
                    labels  = churn[column].value_counts().keys().tolist(),
                    hoverinfo = "label+percent+name",
                    domain  = dict(x = [0,.48]),
                    name    = "Churn Customers",
                    marker  = dict(line = dict(width = 2,
                                               color = "rgb(243,243,243)")
                                  ),
                    hole    = .6
                   )
    trace2 = go.Pie(values  = not_churn[column].value_counts().values.tolist(),
                    labels  = not_churn[column].value_counts().keys().tolist(),
                    hoverinfo = "label+percent+name",
                    marker  = dict(line = dict(width = 2,
                                               color = "rgb(243,243,243)")
                                  ),
                    domain  = dict(x = [.52,1]),
                    hole    = .6,
                    name    = "Non churn customers" 
                   )


    layout = go.Layout(dict(title = column + " distribution in customer Churn ",
                            plot_bgcolor  = "rgb(243,243,243)",
                            paper_bgcolor = "rgb(243,243,243)",
                            annotations = [dict(text = "churn customers",
                                                font = dict(size = 13),
                                                showarrow = False,
                                                x = .15, y = .5),
                                           dict(text = "Non churn customers",
                                                font = dict(size = 13),
                                                showarrow = False,
                                                x = .88,y = .5
                                               )
                                          ]
                           )
                      )
    data = [trace1,trace2]
    fig  = go.Figure(data = data,layout = layout)
    iplot(fig)
In [25]:
# for all categorical columns plot pie

for i in cat_cols :
    plot_pie(i)

products_number affects to churn a lot as there are huge difference between the products which affect to churn. Otherthan that all 4 columns has almost equal ratio of churn and not churn.¶

Varibles distribution in Churn and Non-churn category: Histograms for Numerical variables¶

In [26]:
#function  for histogram for customer attrition types
def histogram(column) :
    trace1 = go.Histogram(x  = churn[column],
                          histnorm= "percent",
                          name = "Churn Customers",
                          marker = dict(line = dict(width = .5,
                                                    color = "black"
                                                    )
                                        ),
                         opacity = .9 
                         ) 
    
    trace2 = go.Histogram(x  = not_churn[column],
                          histnorm = "percent",
                          name = "Non churn customers",
                          marker = dict(line = dict(width = .5,
                                              color = "black"
                                             )
                                 ),
                          opacity = .9
                         )
    
    data = [trace1,trace2]
    layout = go.Layout(dict(title =column + " distribution in customer Churn ",
                            plot_bgcolor  = "rgb(243,243,243)",
                            paper_bgcolor = "rgb(243,243,243)",
                            xaxis = dict(gridcolor = 'rgb(255, 255, 255)',
                                             title = column,
                                             zerolinewidth=1,
                                             ticklen=5,
                                             gridwidth=2
                                            ),
                            yaxis = dict(gridcolor = 'rgb(255, 255, 255)',
                                             title = "percent",
                                             zerolinewidth=1,
                                             ticklen=5,
                                             gridwidth=2
                                            ),
                           )
                      )
    fig  = go.Figure(data=data,layout=layout)
    
    iplot(fig)
In [27]:
#for all categorical columns plot histogram    
for i in num_cols :
    histogram(i)
plt.show()

From the above graph, it can be said that age as well as credit score affect a lot to churn.¶

Scatter Plot for Numerical variables¶

In [28]:
#function  for scatter plot matrix  for numerical columns in data
def scatter_matrix(df)  :
    
    df  = df.sort_values(by = "churn" ,ascending = True)
    classes = df["churn"].unique().tolist()
    classes
    
    class_code  = {classes[k] : k for k in range(2)}
    class_code

    color_vals = [class_code[cl] for cl in df["churn"]]
    color_vals

    pl_colorscale = "Portland"

    pl_colorscale

    text = [df.loc[k,"churn"] for k in range(len(df))]
    text

    trace = go.Splom(dimensions = [dict(label  = "credit_score",
                                       values = df["credit_score"]),
                                  dict(label  = 'age',
                                       values = df['age']),
                                  dict(label  = 'tenure',
                                       values = df['tenure']),
                                  dict(label  = 'balance',
                                       values = df['balance']),
                                  dict(label  = 'estimated_salary',
                                       values = df['estimated_salary'])],
                                  
                     text = text,
                     marker = dict(color = color_vals,
                                   colorscale = pl_colorscale,
                                   size = 3,
                                   showscale = False,
                                   line = dict(width = .1,
                                               color='rgb(230,230,230)'
                                              )
                                  )
                    )
    axis = dict(showline  = True,
                zeroline  = False,
                gridcolor = "#fff",
                ticklen   = 4
               )
    
    layout = go.Layout(dict(title  = 
                            "Scatter plot matrix for Numerical columns for customer churn",
                            autosize = False,
                            height = 800,
                            width  = 800,
                            dragmode = "select",
                            hovermode = "closest",
                            plot_bgcolor  = 'rgba(240,240,240, 0.95)',
                            xaxis1 = dict(axis),
                            yaxis1 = dict(axis),
                            xaxis2 = dict(axis),
                            yaxis2 = dict(axis),
                            xaxis3 = dict(axis),
                            yaxis3 = dict(axis),
                           )
                      )
    data   = [trace]
    fig = go.Figure(data = data,layout = layout )
    iplot(fig)
In [29]:
#scatter plot matrix
scatter_matrix(bank)

Data Pre-Processing¶

In [30]:
from sklearn.preprocessing import LabelEncoder
from sklearn.preprocessing import StandardScaler


#customer id col
Id_col     = ['customer_id']
#Target columns
target_col = ["churn"]
#categorical columns
cat_cols   = bank.nunique()[bank.nunique() < 6].keys().tolist()
cat_cols   = [x for x in cat_cols if x not in target_col]
#numerical columns
num_cols   = [x for x in bank.columns if x not in cat_cols + target_col + Id_col]

#Binary columns with 2 values
bin_cols   = bank.nunique()[bank.nunique() == 2].keys().tolist()
#Columns more than 2 values
multi_cols = [i for i in cat_cols if i not in bin_cols]
In [31]:
multi_cols
Out[31]:
['country', 'products_number']
In [32]:
#Label encoding Binary columns
le = LabelEncoder()
for i in bin_cols :
    bank[i] = le.fit_transform(bank[i])
    
#Duplicating columns for multi value columns
bank = pd.get_dummies(data = bank,columns = multi_cols)
In [33]:
bank
Out[33]:
customer_id credit_score gender age tenure balance credit_card active_member estimated_salary churn country_France country_Germany country_Spain products_number_1 products_number_2 products_number_3 products_number_4
0 15634602 619 0 42 2 0.00 1 1 101348.88 1 1 0 0 1 0 0 0
1 15647311 608 0 41 1 83807.86 0 1 112542.58 0 0 0 1 1 0 0 0
2 15619304 502 0 42 8 159660.80 1 0 113931.57 1 1 0 0 0 0 1 0
3 15701354 699 0 39 1 0.00 0 0 93826.63 0 1 0 0 0 1 0 0
4 15737888 850 0 43 2 125510.82 1 1 79084.10 0 0 0 1 1 0 0 0
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
9995 15606229 771 1 39 5 0.00 1 0 96270.64 0 1 0 0 0 1 0 0
9996 15569892 516 1 35 10 57369.61 1 1 101699.77 0 1 0 0 1 0 0 0
9997 15584532 709 0 36 7 0.00 0 1 42085.58 1 1 0 0 1 0 0 0
9998 15682355 772 1 42 3 75075.31 1 0 92888.52 1 0 1 0 0 1 0 0
9999 15628319 792 0 28 4 130142.79 1 0 38190.78 0 1 0 0 1 0 0 0

10000 rows × 17 columns

In [34]:
#Scaling Numerical columns
std = StandardScaler()
scaled = std.fit_transform(bank[num_cols])
scaled = pd.DataFrame(scaled,columns=num_cols)
In [35]:
scaled
Out[35]:
credit_score age tenure balance estimated_salary
0 -0.326221 0.293517 -1.041760 -1.225848 0.021886
1 -0.440036 0.198164 -1.387538 0.117350 0.216534
2 -1.536794 0.293517 1.032908 1.333053 0.240687
3 0.501521 0.007457 -1.387538 -1.225848 -0.108918
4 2.063884 0.388871 -1.041760 0.785728 -0.365276
... ... ... ... ... ...
9995 1.246488 0.007457 -0.004426 -1.225848 -0.066419
9996 -1.391939 -0.373958 1.724464 -0.306379 0.027988
9997 0.604988 -0.278604 0.687130 -1.225848 -1.008643
9998 1.256835 0.293517 -0.695982 -0.022608 -0.125231
9999 1.463771 -1.041433 -0.350204 0.859965 -1.076370

10000 rows × 5 columns

In [36]:
bank = bank.drop(columns = num_cols,axis = 1)
bank = bank.merge(scaled,left_index=True,right_index=True,how = "left")
In [37]:
bank
Out[37]:
customer_id gender credit_card active_member churn country_France country_Germany country_Spain products_number_1 products_number_2 products_number_3 products_number_4 credit_score age tenure balance estimated_salary
0 15634602 0 1 1 1 1 0 0 1 0 0 0 -0.326221 0.293517 -1.041760 -1.225848 0.021886
1 15647311 0 0 1 0 0 0 1 1 0 0 0 -0.440036 0.198164 -1.387538 0.117350 0.216534
2 15619304 0 1 0 1 1 0 0 0 0 1 0 -1.536794 0.293517 1.032908 1.333053 0.240687
3 15701354 0 0 0 0 1 0 0 0 1 0 0 0.501521 0.007457 -1.387538 -1.225848 -0.108918
4 15737888 0 1 1 0 0 0 1 1 0 0 0 2.063884 0.388871 -1.041760 0.785728 -0.365276
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
9995 15606229 1 1 0 0 1 0 0 0 1 0 0 1.246488 0.007457 -0.004426 -1.225848 -0.066419
9996 15569892 1 1 1 0 1 0 0 1 0 0 0 -1.391939 -0.373958 1.724464 -0.306379 0.027988
9997 15584532 0 0 1 1 1 0 0 1 0 0 0 0.604988 -0.278604 0.687130 -1.225848 -1.008643
9998 15682355 1 1 0 1 0 1 0 0 1 0 0 1.256835 0.293517 -0.695982 -0.022608 -0.125231
9999 15628319 0 1 0 0 1 0 0 1 0 0 0 1.463771 -1.041433 -0.350204 0.859965 -1.076370

10000 rows × 17 columns

In [38]:
summary = bank.describe().transpose()
summary
Out[38]:
count mean std min 25% 50% 75% max
customer_id 10000.0 1.569094e+07 71936.186123 1.556570e+07 1.562853e+07 1.569074e+07 1.575323e+07 1.581569e+07
gender 10000.0 5.457000e-01 0.497932 0.000000e+00 0.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00
credit_card 10000.0 7.055000e-01 0.455840 0.000000e+00 0.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00
active_member 10000.0 5.151000e-01 0.499797 0.000000e+00 0.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00
churn 10000.0 2.037000e-01 0.402769 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 1.000000e+00
country_France 10000.0 5.014000e-01 0.500023 0.000000e+00 0.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00
country_Germany 10000.0 2.509000e-01 0.433553 0.000000e+00 0.000000e+00 0.000000e+00 1.000000e+00 1.000000e+00
country_Spain 10000.0 2.477000e-01 0.431698 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 1.000000e+00
products_number_1 10000.0 5.084000e-01 0.499954 0.000000e+00 0.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00
products_number_2 10000.0 4.590000e-01 0.498341 0.000000e+00 0.000000e+00 0.000000e+00 1.000000e+00 1.000000e+00
products_number_3 10000.0 2.660000e-02 0.160919 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 1.000000e+00
products_number_4 10000.0 6.000000e-03 0.077231 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 1.000000e+00
credit_score 10000.0 -4.824585e-16 1.000050 -3.109504e+00 -6.883586e-01 1.522218e-02 6.981094e-01 2.063884e+00
age 10000.0 2.318146e-16 1.000050 -1.994969e+00 -6.600185e-01 -1.832505e-01 4.842246e-01 5.061197e+00
tenure 10000.0 -1.078249e-16 1.000050 -1.733315e+00 -6.959818e-01 -4.425957e-03 6.871299e-01 1.724464e+00
balance 10000.0 -6.252776e-17 1.000050 -1.225848e+00 -1.225848e+00 3.319639e-01 8.199205e-01 2.795323e+00
estimated_salary 10000.0 -2.877698e-17 1.000050 -1.740268e+00 -8.535935e-01 1.802807e-03 8.572431e-01 1.737200e+00

Variable summary¶

In [39]:
# To make it look nice we can do some additional stuff if needed
summary = (bank[[i for i in bank.columns if i not in Id_col]].
           describe().transpose().reset_index())

summary = summary.rename(columns = {"index" : "feature"})
summary = np.around(summary,3)

val_lst = [summary['feature'], summary['count'],
           summary['mean'],summary['std'],
           summary['min'], summary['25%'],
           summary['50%'], summary['75%'], summary['max']]

trace  = go.Table(header = dict(values = summary.columns.tolist(),
                                line = dict(color = ['#506784']),
                                fill = dict(color = ['#119DFF']),
                               ),
                  cells  = dict(values = val_lst,
                                line = dict(color = ['#506784']),
                                fill = dict(color = ["lightgrey",'#F5F8FF'])
                               ),
                  columnwidth = [200,60,100,100,60,60,80,80,80])
layout = go.Layout(dict(title = "Variable Summary"))
figure = go.Figure(data=[trace],layout=layout)
iplot(figure)
In [40]:
#correlation
correlation = bank.corr()
#tick labels
matrix_cols = correlation.columns.tolist()
In [41]:
#convert to array
corr_array  = np.array(correlation)
In [42]:
correlation
Out[42]:
customer_id gender credit_card active_member churn country_France country_Germany country_Spain products_number_1 products_number_2 products_number_3 products_number_4 credit_score age tenure balance estimated_salary
customer_id 1.000000 -0.002641 -0.014025 0.001665 -0.006248 -0.004049 -0.003097 0.007800 -0.018501 0.014917 0.018706 -0.015462 0.005308 0.009497 -0.014883 -0.012419 0.015271
gender -0.002641 1.000000 0.005766 0.022544 -0.106512 0.006772 -0.024628 0.016889 0.005488 0.010171 -0.035143 -0.027936 -0.002857 -0.027544 0.014733 0.012087 -0.008112
credit_card -0.014025 0.005766 1.000000 -0.011866 -0.007138 0.002467 0.010577 -0.013480 -0.003845 0.003414 0.003186 -0.003778 -0.005458 -0.011721 0.022583 -0.014858 -0.009933
active_member 0.001665 0.022544 -0.011866 1.000000 -0.156128 0.003317 -0.020486 0.016732 -0.022321 0.032802 -0.029864 -0.004938 0.025651 0.085472 -0.028362 -0.010084 -0.011421
churn -0.006248 -0.106512 -0.007138 -0.156128 1.000000 -0.104955 0.173488 -0.052667 0.185447 -0.292474 0.255862 0.153612 -0.027094 0.285323 -0.014001 0.118533 0.012097
country_France -0.004049 0.006772 0.002467 0.003317 -0.104955 1.000000 -0.580359 -0.575418 -0.014049 0.026318 -0.036508 -0.002807 -0.008928 -0.039208 -0.002848 -0.231329 -0.003332
country_Germany -0.003097 -0.024628 0.010577 -0.020486 0.173488 -0.580359 1.000000 -0.332084 0.033878 -0.051673 0.041945 0.026720 0.005538 0.046897 -0.000567 0.401110 0.010297
country_Spain 0.007800 0.016889 -0.013480 0.016732 -0.052667 -0.575418 -0.332084 1.000000 -0.017750 0.021411 0.000161 -0.023583 0.004780 -0.001685 0.003868 -0.134892 -0.006482
products_number_1 -0.018501 0.005488 -0.003845 -0.022321 0.185447 -0.014049 0.033878 -0.017750 1.000000 -0.936709 -0.168109 -0.079009 -0.014813 0.072852 -0.013423 0.359646 -0.010663
products_number_2 0.014917 0.010171 0.003414 0.032802 -0.292474 0.026318 -0.051673 0.021411 -0.936709 1.000000 -0.152266 -0.071563 0.015819 -0.102604 0.012438 -0.363260 0.005795
products_number_3 0.018706 -0.035143 0.003186 -0.029864 0.255862 -0.036508 0.041945 0.000161 -0.168109 -0.152266 1.000000 -0.012843 -0.004145 0.067365 -0.000517 -0.002722 0.012153
products_number_4 -0.015462 -0.027936 -0.003778 -0.004938 0.153612 -0.002807 0.026720 -0.023583 -0.079009 -0.071563 -0.012843 1.000000 0.002455 0.050092 0.007716 0.021476 0.006314
credit_score 0.005308 -0.002857 -0.005458 0.025651 -0.027094 -0.008928 0.005538 0.004780 -0.014813 0.015819 -0.004145 0.002455 1.000000 -0.003965 0.000842 0.006268 -0.001384
age 0.009497 -0.027544 -0.011721 0.085472 0.285323 -0.039208 0.046897 -0.001685 0.072852 -0.102604 0.067365 0.050092 -0.003965 1.000000 -0.009997 0.028308 -0.007201
tenure -0.014883 0.014733 0.022583 -0.028362 -0.014001 -0.002848 -0.000567 0.003868 -0.013423 0.012438 -0.000517 0.007716 0.000842 -0.009997 1.000000 -0.012254 0.007784
balance -0.012419 0.012087 -0.014858 -0.010084 0.118533 -0.231329 0.401110 -0.134892 0.359646 -0.363260 -0.002722 0.021476 0.006268 0.028308 -0.012254 1.000000 0.012797
estimated_salary 0.015271 -0.008112 -0.009933 -0.011421 0.012097 -0.003332 0.010297 -0.006482 -0.010663 0.005795 0.012153 0.006314 -0.001384 -0.007201 0.007784 0.012797 1.000000
In [43]:
#Plotting
trace = go.Heatmap(z = corr_array,
                   x = matrix_cols,
                   y = matrix_cols,
                   colorscale = "Viridis",
                   colorbar   = dict(title = "Pearson Correlation coefficient",
                                     titleside = "right"
                                    ) ,
                  )

layout = go.Layout(dict(title = "Correlation Matrix for variables",
                        autosize = False,
                        height  = 720,
                        width   = 800,
                        margin  = dict(r = 0 ,l = 210,
                                       t = 25,b = 210,
                                      ),
                        yaxis   = dict(tickfont = dict(size = 9)),
                        xaxis   = dict(tickfont = dict(size = 9))
                       )
                  )

data = [trace]
fig = go.Figure(data=data,layout=layout)
iplot(fig)

Model Building (We will build Decision Tree and Logistics Regression models)¶

In [44]:
# Import all the modules

from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.tree import DecisionTreeClassifier
from sklearn.metrics import confusion_matrix,accuracy_score,classification_report
from sklearn.metrics import f1_score
import statsmodels.api as sm
from sklearn.metrics import precision_score,recall_score
In [45]:
#splitting train and test data 
train,test = train_test_split(bank,test_size = .25 ,random_state = 111)
In [46]:
##seperating dependent and independent variables
cols    = [i for i in bank.columns if i not in Id_col + target_col]
X_train = train[cols]
Y_train = train[target_col]
X_test  = test[cols]
Y_test  = test[target_col]

Logistics Regression¶

In [47]:
classifier = LogisticRegression(random_state = 0)
classifier.fit(X_train, Y_train)
Out[47]:
LogisticRegression(random_state=0)
In [48]:
# Predicting test set
y_pred = classifier.predict(X_test)
In [49]:
classifier.coef_
Out[49]:
array([[-0.5790545 , -0.07088232, -1.09303953, -0.3507228 ,  0.63168598,
        -0.28143816, -1.09019395, -2.66869861,  1.31184263,  2.44657494,
        -0.03056978,  0.73546963, -0.06948406, -0.05155216,  0.0042424 ]])
In [50]:
#Evaluating the Results
cm = confusion_matrix(Y_test, y_pred)
cm
Out[50]:
array([[1900,   82],
       [ 337,  181]], dtype=int64)
In [51]:
accuracy_score(Y_test, y_pred)
Out[51]:
0.8324
In [52]:
precision_score(Y_test, y_pred)
Out[52]:
0.688212927756654
In [53]:
f1_score(Y_test, y_pred)
Out[53]:
0.4635083226632522
In [54]:
recall_score(Y_test, y_pred)
Out[54]:
0.34942084942084944
In [55]:
df_cm = pd.DataFrame(cm, index = (0, 1), columns = (0, 1))
plt.figure(figsize = (7,5))
sns.heatmap(df_cm, annot = True, fmt ='g')
print("Test Data Accuracy: %.4f" %accuracy_score(Y_test, y_pred))

Test Data Accuracy: 0.8324

Decision Tree¶

In [56]:
model_tree = DecisionTreeClassifier(random_state = 2)
model_tree.fit(X_train, Y_train)
Out[56]:
DecisionTreeClassifier(random_state=2)
In [57]:
# Predicting test set
y_pred = model_tree.predict(X_test)
In [58]:
#Evaluating the Results
cm = confusion_matrix(Y_test, y_pred)
cm
Out[58]:
array([[1733,  249],
       [ 275,  243]], dtype=int64)
In [59]:
accuracy_score(Y_test, y_pred)
Out[59]:
0.7904
In [60]:
precision_score(Y_test, y_pred)
Out[60]:
0.49390243902439024
In [61]:
f1_score(Y_test, y_pred)
Out[61]:
0.48118811881188117
In [62]:
recall_score(Y_test, y_pred)
Out[62]:
0.4691119691119691
In [63]:
df_cm = pd.DataFrame(cm, index = (0, 1), columns = (0, 1))
plt.figure(figsize = (7,5))
sns.heatmap(df_cm, annot = True, fmt ='g')
print("Test Data Accuracy: %.4f" %accuracy_score(Y_test, y_pred))

Test Data Accuracy: 0.7904

Feature Seclection/Feature Engineering¶

In [64]:
X = bank.drop("churn", axis=1)
y = bank["churn"]
from sklearn.ensemble import ExtraTreesClassifier
import matplotlib.pyplot as plt
model = ExtraTreesClassifier()
model.fit(X,y)
print(model.feature_importances_) 
 
#plot the graph of feature importances 
feat_importances = pd.Series(model.feature_importances_, index=X.columns)
feat_importances.nlargest(10).plot(kind='barh')
plt.show()

[0.10718613 0.0183658  0.02250042 0.03858461 0.00910023 0.02382487
 0.00854272 0.03507846 0.06051981 0.03957946 0.01217415 0.10832772
 0.20034353 0.09134625 0.11627307 0.10825276]
In [65]:
final_results = pd.concat([Y_test, bank.customer_id], axis =1).dropna()
final_results['predicted_churn'] = y_pred
final_results = final_results[['customer_id', 'churn', 'predicted_churn']].reset_index(drop = True)
In [66]:
final_results
Out[66]:
customer_id churn predicted_churn
0 15634602 1.0 0
1 15619304 1.0 0
2 15737452 1.0 0
3 15577657 0.0 0
4 15738191 0.0 0
... ... ... ...
2495 15666295 1.0 0
2496 15656710 0.0 1
2497 15586914 0.0 1
2498 15769959 1.0 0
2499 15584532 1.0 0

2500 rows × 3 columns

There are two models created namely logistic regression and decision tree to predict the probability of customer churning.¶

From the result, it can be said that age is most significant feature, that means, older customers are more likely to churn. Apart from that number of product has a high impact on churning.¶

The bank may come up with some strong policies to engage with elderly people so that churning can be reduced. Also the bank maycome up with a very strong rewarding programmes as weel as attractive loan interest.¶

In [ ]: